Technical Comments

Comment on “Single-Crystal X-ray Structure of 1,3-Dimethylcyclobutadiene by Confinement in a Crystalline Matrix”

See allHide authors and affiliations

Science  19 Nov 2010:
Vol. 330, Issue 6007, pp. 1047
DOI: 10.1126/science.1196188


Legrand et al. (Reports, 16 July 2010, p. 299) reported the experimental observation of square-planar and rectangular-bent geometries of 1,3-dimethylcyclobutadiene (Me2CBD) confined within a crystalline matrix. However, we found no evidence for the Me2CBD formation. We argue that the experimental x-ray density data are better attributed to the bicyclic β-lactone intermediate where carbon dioxide is covalently bound to cyclobutadiene.

Cyclobutadiene (CBD), the smallest neutral example of Hückel anti-aromaticity, has long been predicted to be a highly reactive species. Adding to the electronic instability is the ring strain created by the large deviation from the usual sp2-trigonal geometry imposed by the four-membered ring. The shape and ground-state electron configuration of CBD were much debated until theoretical predictions converged on a rectangular singlet ground state rather than the Hückel’s triplet, D4h structure (13).

The challenge provided by anti-aromatic character and intrinsic instability of CBD inspired numerous experimental studies (48). The photochemical transformation of α-pyrone (9, 10) (Scheme 1) confined within a matrix emerged as a promising route to this elusive molecule without protection provided by bulky substituents or transition metals (11).

Scheme 1

Photochemical preparation of CBD derivatives from α-pyrone.

Legrand et al. (12) redesigned Cram’s approach (13) for trapping CBD in a hemicarcerand cage to the preparation of 1,3-dimethylcyclobutadiene (Me2CBD) from 4,6-dimethyl-α-pyrone immobilized in a crystalline cavity of a host matrix at 175 K. Based on x-ray diffraction data, the authors suggested that the guest undergoes initial photochemical transformation into a 4,6-dimethyl-β-lactone Dewar intermediate (Scheme 1). Further irradiation was reported to push the reaction to completion via loss of CO2 from the intermediate, thus “enabling the structure determination of 1,3-dimethylcyclobutadiene.”

An exciting ramification is that these results would reveal direct x-ray structural information regarding CBD, “the Mona Lisa of organic chemistry” (13). Indeed, Legrand et al. reported the experimental observation of square-planar and rectangular-bent geometries in the host matrix (Fig. 1). The main part of the diffraction data (62.7%) for Me2CBD was assigned to the square geometry (Me2CBDS), with the bent-rectangular form (Me2CBDR) being a minor component (37.3%). These preferences are opposite to theoretical predictions for an isolated CBD molecule (13), where the square form is expected to be a transition state between two Jahn-Teller distorted rectangles.

Fig. 1

Geometries of square (Me2CDBS) and rectangular-bent (Me2CBDR) dimethylcyclobutadiene from (12).

We were intrigued by this interpretation but identified a number of experimental inconsistencies. Initially, we were troubled by the unusually long (1.70 Å) C3–C6 bridge bond in the structure assigned to the bicyclic lactone. This value is only marginally lower than the longest known C–C bond (1.72 Å) (14, 15) and deviates substantially from the theoretical value [1.55 Å, according to our density functional theory (DFT) calculations (16)]. There are large differences for other C–C bonds as well, e.g., dC2-C3 = dC3-C4 = 1.37 Å versus 1.53 Å in theory. Even considering the effect imposed by the solid-state environment, the deviations are large enough to raise a concern about the accuracy of the results presented by Legrand et al. (12).

A stunning discrepancy, overlooked by Legrand et al. but obvious upon inspection of figure 4 in (12) (reproduced partially in Fig. 1 here) is that the “bystander” CO2 molecule present in the cavity is strongly bent (119.9°)—a striking deviation from the textbook linear geometry but a good match for a trigonal planar carbon. According to the DFT calculations (16), such bending would cost ~75 kcal/mol, even without any additional structural distortions. Considering that one of the two C–O bonds in the reported x-ray geometry was shorter than the other (1.14 Å versus 1.34 Å), the overall penalty for the structural distortion from the optimal linear CO2 increases to ~82 kcal/mol, a thermodynamic cost comparable in energy to a chemical bond.

Equally noteworthy are the very short distances between the CO2 molecule and Me2CBD: 1.50 and 1.61 Å. Legrand et al. attribute this finding to a “strong van der Waals contact” rather than to covalent bonding between the confined CO2 and Me2CBDR molecules. This interpretation is questionable because the purported 1.50 Å C2–C3 van der Waals contact between Me2CBDS and CO2 is shorter than the 1.56 Å C–C bond in Me2CBDR. Can a van der Waals contact really be shorter than a bond? The tabulated nonbonding van der Waals C–C and C–O contacts are 3.50 Å and 3.24 Å, respectively (17), both of them being substantially longer than the aforementioned “strong van der Waals” contacts.

On reexamination of the x-ray data of Legrand et al. (12), we found that the crystallographic results and their quality raise serious concerns about the conclusions drawn from them. A detailed assessment of the crystallographic work is given in the Supporting Online Material. Here, we will focus on the interpretation of structural parameters.

Figure 2 depicts the proposed bicyclic intermediate, along with alternative views of two components of the disordered Me2CBD molecule, labeled according to Legrand et al. (12). It is immediately obvious that the C2–C3 bond length of 1.50(2) Å in Me2CBDS is comparable to those within the CBD unit. In Me2CBDR, this distance is even shorter, 1.41(3) Å, and three of the bonds in the CBD unit are longer than this value. The CO2 and CBD units are thus almost certainly connected by a covalent C–C bond. In fact, the experimental x-ray density of Me2CBDS is better attributed to CO2 covalently bound to CBD in a structure similar to the theoretically predicted geometry of the Dewar intermediate.

Fig. 2

Two components of crystallographic disorder of the guest molecule in the crystal structure of Me2CBD (top) and theoretically calculated and experimentally suggested structure of the purported bicyclic intermediate (bottom). The bond lengths in Me2CBDS (top left) are very similar to those predicted theoretically for the Dewar intermediate (bottom left).

Why did CBD not form? We offer two possibilities: (i) free volume restrictions in the carcerand prevent the fragmentation or (ii) the bicyclic lactone is transparent in the 320- to 500-nm excitation range employed by the authors. It is known that the β-lactone intermediate only gives rise to CO2 and CBD when irradiated with photons of much higher energy (18) (Scheme 1). Interestingly, even in those cases, CO2 is known to affect spectral properties of “free” CBD (5, 19). Regrettably, we thus conclude that the crystallographic analysis of Me2CBD remains an unsolved experimental challenge. The “Mona Lisa of organic chemistry” still smiles at us but keeps her secret.

Supporting Online Material

SOM Text

Table S1

References and Notes

  1. In the actual crystallographic information file provided as supporting information by Legrand et al. (12), this bond is even longer, 1.79(3) Å. Thus, there is also an obvious discrepancy between the values given in the text and those found in the crystal structure of the purported Dewar intermediate.
  2. DFT calculations were performed at the B3LYP/6-31G** level of theory.
  3. I.V.A. is funded in part by the National Science Foundation (CHE-0848686) and Petroleum Research Fund, administered by the American Chemical Society (Award 47590-AC4).
View Abstract

Navigate This Article